Supercomputing Power in Gaming Graphics Accelerators

Today’s video games use computing power that was achieved only by supercomputers just a few years ago and the makers of computer graphics cards have risen to the challenge. Screen resolution continues to climb and the visual effects get bigger and better.

Nvidia Tesla graphics card
AMD’s Radeon series and Nvidia’s GeForce series graphics accelerators achieve the effects by dividing the computations between hundreds, or even thousands of GPUs or Graphics Processing Units.

Nvidia’s graphics chips also process the physics equations involved in explosions, shock waves and ricochets.

While the performance of the main processors in computers continues to improve, the speed of the GPUs is now several times faster. As a result, when the US government decided to build the world’s fastest supercomputer, they recently chose a design that incorporates thousands of Nvidia GPUs.

The Department of Energy just signed a deal with IBM and Nvidia to build two supercomputers for the Oak Ridge National Laboratory and the Lawrence Livermore National Laboratory featuring thousands of IBM servers and even more Nvidia graphics accelerators based on the Tesla platform pictured above. The total value of the deal is with $325 million, far above the average video gamer’s budget. Nvidia further accelerates the processing by stacking the RAM memory on the GPUs. The new supercomputer named “Summit” at Oak Ridge will be capable of 150 to 300 petaflops (quadrillion floating point calculations per second).

The cutting-edge same technology is not only available to video game players, but it may be coming to an automobile near you. Nvidia has partnered with automakers worldwide including Audi, Honda, Tesla and Volkswagen for infotainment systems, navigation and collision avoidance systems.

Leave a Reply

Your email address will not be published. Required fields are marked *